A computer system not only may face a problem, such as an operational failure or a performance failure, but also may need to change the system setting when the setting does not match with the purpose of the computer system. The occurrence of such a problem or the need to change the setting is called “symptom”. Detecting and attending to a symptom can ensure efficient and safe management of a computer system. Because recent computer systems are complex, however, it is not easy to detect and attend to a symptom when occurred therein.
There has been proposed a technique of allowing a computer to detect and attend to a symptom occurred therein (see “Blueprint on autonomic computing architecture” URL:http://www-06.ibm.com/jp/autonomic/pdf/2006_AC_Blueprint—2006—06.pdf)
In autonomic computing, a system of detecting and attending to symptoms is called an autonomic manager, which includes analysis, plan and execution functions (see pp. 10-11 of the above article). To add a new function to the autonomic manager to be able to, for example, detect a new symptom, a knowledge base is used (see p. 12 of the above article). To adequately function the system, therefore, it is desirable to sufficiently collect information of events needed for analysis first, then permit the use of a sufficient processing capability to achieve a plan and execution thereof, and further to provide a sufficient storage area for storing the knowledge base.
Recent devices, such as cellular phones, PDAs (Personal Data Assistants) and home appliances, are equipped with various functions of a computer. However, those devices do not have a sufficient processing capability or a sufficient storage area, thus making it difficult to allow those devices to adequately operate an autonomic manager. It is also difficult to detect occurrence of a specific symptom from the operation of a device itself. Such a specific symptom may occur due to complex factors including a server which communicates with the device. Even with a sufficient processing capability, therefore, it may not be proper for each device to operate an autonomic manager.
Because such devices are becoming exceedingly popular recently, a server may not be able to intensively manage symptoms occurred in individual devices due to the insufficient processing capability of the server. Those devices often suffer an unstable communication state, so that the server may not adequately collect the statuses of the individual devices even if it tries. Further, a system which allows each device to request the server to detect and attend to a symptom takes a longer process time by the time needed for a communication process, as compared with a case where a device itself detects and attends to a symptom. This is likely to reduce the operability and convenience.